Le 12/12/2012 18:26, Ran Atkinson a écrit :
On 12 Dec 2012, at 12:21 , Rémi Després wrote:
Do you know any deployment with u=g=1 in this context ?
There is limited use today of IPv6 addresses that have apparent
unicast routing prefixes, but contain multicast group IDs in the
low-order 64-bits ("IID").
Interesting to know, thanks.
But, this wouldn't make the ug(ly) bits mandatory in unicast, would it.
In addition, together with the ug bits, the 64bit fixed length of the
IID (regardless of Ethernet) is another factor posing problems to
deployments (e.g. ISP delivering /64 to end user, and end-user doing
non-natural things like duplicating same address on two interfaces, or
alternatively hopelessly waiting of DHCPv6-PD deployment on otherwise
IPv6 ptp cellular links available now).
If the ug bits didnt exist, and if the 64bit IID length were not fixed,
then we could have happy ISPs delivering /64 on ptp links that end user
would naturally split ('subnet') into /65 and longer without wait, and
without need of DHCPv6 on these /65 links (i.e. use RA SLAAC with /65 on
Ethernet).
YOurs,
Alex
Yours,
Ran
--------------------------------------------------------------------
IETF IPv6 working group mailing list ipv6@ietf.org Administrative
Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
